TOME VI (year 008) FASCICULE (ISSN 1584 665) PERFORMANCE-BASED DESIGN APPLIED FOR A BEAM SUBJECTED TO COMBINED STRESS Karel FRYDRÝŠEK Department of Mechanics of Materials Faculty of Mechanical Engineering VŠB-Technical University of Ostrava CZECH REPUBLIC ABSTRACT: Performance-Based Design (PBD) is based on the theory of probability connected with statistics. PBD is also based on the performance requirements (PR) which are usually defined as a synthesis of functionality all-in cost safety etc. PR can be expressed as an acceptable level of damage (i.e. acceptable probability of possible failure). In presented example (i.e. a shaft of unknown circular shape is exposed to bending moment normal force and torque which are given by bounded histograms) is used Simulation-Based Reliability Assessment Method method (direct Monte Carlo Method AntHill software). The task is to calculate the nominal value of diameter which is given by normal bounded distribution. The acceptable level of damage is related to the yield stress. The calculation of the diameter (i.e. solution of the inverse problem of theory of probability) must be solved via iterative approaches. To get the solution of this type of inverse problem is much difficult than the solution of the classical problem of theory of probability. KEYWORDS: beam combined loading probability analysis Simulation-Based Reliability Assessment (SBRA) method Monte Carlo simulations Performance-Based Design 1. INTRODUCTIVE NOTES In the last several decades the science and engineering community has progressively ventured outside of traditional boundaries in terms of materials loads configurations etc. for structural systems in mechanics. Consequently a new designer's approach called Performance-Based Design (PBD) can be defined as: Design specifically intended to limit the consequences of one or more perils to defined acceptable levels. PBD is based on the theory of probability and depends on many inter-connected issues including classification of constructed systems definition of performance tools for measuring performance quantitative indices that may serve as assurance of performance and especially how to describe and measure performance especially under various levels of uncertainty which is connected with statistics. However comprehensive approach of PBD is still in its infancy. PBD is based on performance requirements which are usually defined as a synthesis of functionality all-in cost safety etc. see Fig.1 and. Performance requirements can be expressed as an acceptable level of damage which is defined by acceptable probability of possible failure P ACCEPT. For more details see reference (Hamburger 1999). 19
ENGINEERING. TOME VI (year 008). Fascicule (ISSN 1584 665) FIGURE 1. STRUCTURE OF PBD FIGURE. STRUCTURE OF PBD.. GIVEN EXAMPLE - SHAFT SUBJECTED TO COMBINED STRESS A shaft of unknown circular diameter D (see Fig.3) is exposed to bending moment 581515.3 8056.6 M Nmm normal force N 157493.7 1777.8 N and torque FIGURE 3. SHAFT WITH UNKNOWN DIAMETER IS SUBJECTED TO COMBINED LOADING of damage is P stress. In other words o 344596.4 M k 9338. 367485.4 506981.6 17765.8 Nmm which are given by truncated histograms see Fig.4 to 6. Yield stress of material is 338.3 161.7 MPa see truncated R e 90.3 histogram in Fig.7. Calculate the value of diameter D which is given by normal 1% truncated distribution ±1% (i.e. ) with D 1% accuracy 0.1mm. The acceptable level 0.0005 0.05% (standard reliability level) is related to yield 0.05 % of all loading states can result in yielding. ACCEPT FIGURE 4. HISTOGRAM OF BENDING MOMENT M o FIGURE 5. HISTOGRAM OF NORMAL FORCE N. FIGURE 6. HISTOGRAM OF TORQUE Mk. FIGURE 7. HISTOGRAM OF YIELD STRESS Re. In the following example is used SBRA method (Simulation-Based Reliability Assessment direct Monte-Carlo method AntHill software) see [] and [3]. 130
ENGINEERING. TOME VI (year 008). Fascicule (ISSN 1584 665) where 3. SOLVED EXAMPLE - PBD APPLIED FOR A SHAFT SUBJECTED TO COMBINED STRESS According to the theory of small deformations (see [4]) can be written: 4N 3Mo 4 8M σ o 16M k N τ (1) 3 D 3 σ /MPa/ is maximal normal stress and Hence for equivalent von Mises stress HMH 4 σ 3τ τ /MPa/ is maximal shear stress. σ HMH /MPa/ can be written: 8M 48 o M k N D D σ. () Factor of safety (i.e. probability of situations when R e < σhmh ) is defined as: FS P( R e σ HMH < 0) (3) where operator P means probability. Hence when FS 0 it is evident that yield limit is not reached (i.e. in the shaft are not any plastic deformations). The goal is to calculate diameter D which satisfy condition: FS P ACCEPT. (4) However it is necessary to applied iteration methods because from eq. () is not possible to express directly the unknown parameter D. Hence iteration loop with application of secant method can be used see Fig.8. FIGURE 8. SECANT METHOD (calculation of D ) FIGURE 9. BISECTION METHOD (calculation of D3 ) For chosen initial conditions (diameters): D 0 30 ± 0.3 mm and D 1 48 ± 0.48 mm 6 (truncated normal distributions ±1% ) is possible to calculate (via SBRA method for 10 Monte Carlo simulations) the values of FS0 and FS1: D 0 30 ± 0.3 mm FS0 0.86999 PACCEPT D 1 48 ± 0.48 mm FS1 0.00003 PACCEPT. Hence FS0 P ACCEPT and FS1 PACCEPT. It is evident that the required diameter D must be in interval: D ( D0 ; D1) ( 30 ± 0.3; 48 ± 0.48) mm. From Fig.8 can be derived new approximation of diameter (i.e. D ) via secant method: D1 ( ) ( PACCEPT FS0 ) D0 ( FS1 PACCEPT ) D f PACCEPT 47.9 ± 0.48 mm. FS1 FS0 From the results of AntHill software follows: D 47.9 ± 0.48 mm FS 0. 000035 PACCEPT D ( D0 ; D ) ( 30 ± 0.3; 47.9 ± 0.48) mm. Next approximation of D (i.e. ) can be also calculated via bisection method see Fig.9. Hence: D0 D D3 38.95 ± 0.39 mm FS3 0.048731 PACCEPT D 3 131
ENGINEERING. TOME VI (year 008). Fascicule (ISSN 1584 665) ( D ; D ) ( 38.95 ± 0.39; 47.90 0.48) mm D 3 ±. Next applications of bisection method give: D3 D D4 43.4 ± 0.43 mm FS4 0. 00057 PACCEPT D ( D4 ; D ) ( 43.4 ± 0.43; 47.90 ± 0.48) mm D4 D D5 45.66 ± 0.46 mm FS5 0. 0009 PACCEPT D ( D4 ; D5 ) ( 43.4 ± 0.43; 45.66 ± 0.46) mm D4 D5 D6 44.54 ± 0.45 mm FS6 0.00080 PACCEPT D ( D6 ; D5 ) ( 44.54 ± 0.45; 45.66 ± 0.46) mm. Because the values of FS5 and FS6 are very close to given value of PACCEPT it 6 is wise to increase the number of Monte Carlo simulations to 3 10. Hence the calculated values of FS will be more accurate. Next application of bisection method gives: D6 D5 D7 45.1± 0.45 mm FS7 0.000499 PACCEPT D ( D6 ; D7 ) ( 44.54 ± 0.45; 45.1± 0.45) mm. Next application of secant method gives: D7 ( ) ( PACCEPT FS6 ) D6 ( FS7 PACCEPT ) D8 f PACCEPT 45.098 ± 0.45 mm FS1 FS6 FS8 0.000544 P DOV D 8 45 ± 0.45 mm D ( D8 ; D7 ) ( 45.0 ± 0.45; 45.1± 0.45) mm. Because D8 D 7 and FS7 PACCEPT. (with defined accuracy 0.1 mm) the diameter is: D D7 45.1± 0.45 mm see histograms shown in Fig.10 and 11. FIGURE 10. FINAL DIAMETER. D 45.1± 0.45 mm. FIGURE 11. FACTOR OF SAFETY FS CALCULATED Hence the diameter of damage P 0.0005 0.05%. ACCEPT FOR D 45.1± 0.45 mm. D 45.1± 0.45 mm is calculated with the acceptable level The results of the presented iteration loop as a function f( FS) Fig.1. Histogram of calculated stress Re HMH v.s. σ are presented in Fig.13 and 14. HMH σ 166.89 59.90 99.31 D is shown in MPa and D histogram of 13
ENGINEERING. TOME VI (year 008). Fascicule (ISSN 1584 665) FIGURE 1. DIAGRAM OF CONVERGENCE FOR CALCULATED DIAMETER ITERATIVE PROCEDURES) D (RESULTS OF FIGURE 13. MAXIMUM STRESS σ HMH (HISTOGRAM FIGURE 14. D HISTOGRAM R e f( σ HMH ) CALCULATED FOR D 45.1± 0.45 mm ). CALCULATED FOR D 45.1± 0.45 mm. 4. CONCLUSIONS Performance-Based Design as a new and modern trend in mechanics is based on theory of probability and stochastic methods. The calculation of the diameter D with given acceptable probability of damage level (i.e. solution of the inverse problem of theory of probability) is P ACCEPT solved via iterative approaches (secant method bisection method). Hence to get the solution of this type of inverse problem is much difficult than the classical problem 7 of theory of probability (iterative approach with usually more than 10 simulations. Instead of secant method or bisection method can be used also another methods such as Regula-Falsi Method etc. The whole iterative procedures and Monte Carlo simulations can be speed-up by application of parallel computers. However on the present days it is impossible to solve the large problems of mechanics via PBD. The reason of this is the low rate of present-day computers. Another applications of SBRA method are presented in [] [3] and [5]. This work has been supported by the Czech project FRVŠ 534/008 F1b. 133
ENGINEERING. TOME VI (year 008). Fascicule (ISSN 1584 665) REFERENCES / BIBLIOGRAPHY [1] Hamburger R.O.: The Challenge of Performance-Based Design http://peer.berkeley.edu [] Marek P. Guštar M. Anagnos T. Simulation-Based Reliability Assessment for Structural Engineers CRC Press Inc. Boca Raton Florida USA 1995 pp.365. [3] Marek P. Brozzetti J. Guštar M. Probabilistic Assessment of Structures Using Monte Carlo Simulation Background Exercises and Software ITAM CAS Prague 003 Czech Republic pp.471. [4] Frydrýšek K. Adámková L.: Mechanics of Materials 1 (Introduction Simple Stress and Strain Basic of Bending) Faculty of Mechanical Engineering VŠB-Technical University of Ostrava Ostrava ISBN 978-80-48-1550-3 Ostrava 007 Czech Republic pp. 179. [5] Frydrýšek K.: Reliability Analysis of Beam on Elastic Nonlinear Foundation In: Applied and Computational Mechanics vol. 1 no. 007 ISSN 180-680X University of West Bohemia Plzeň Czech Republic pp. 445-45. 134